Finding relevant features to characterize student behavior on an e-learning system

E-learning systems such as virtual campus environments have established themselves as a strong alternative to traditional distance university education. In this context, Internet allows the gathering of plenty of information on students' online behavior. The knowledge extracted from this information can be used to define personalization strategies tailored to the students' needs and requirements. In this brief study we introduce a model to assess the relative relevance of individual data features on the grouping structure of the users of a real virtual campus (Open University of Catalonia). Tests carried out on these data indicate that certain web navigation behaviors are best at explaining and discriminating the different typologies of online students.